Gage R&R (gauge repeatability and reproducibility) is a measurement system study that shows how much of the variation in your test results comes from the test system itself rather than from the parts. For automated test equipment, it answers a simple question: can this tester reliably tell a good unit from a bad one, close to the limits, on every station?
This article is part of our complete guide to automated test equipment (ATE). Here we focus on how to run a gage R&R on an automated tester, how to read the results, and how to keep a test system capable over its life.
Why gage R&R matters for automated test
Every measurement has some noise. If that noise is small compared with your tolerance, the tester sorts parts correctly. If it is large, units near the limits pass on one run and fail on the next. The result is false failures that waste good product and retest time, and escapes that reach your customer.
Automated testers remove most operator influence, so it is tempting to assume they are capable by default. They are not. Fixture contact, instrument drift, cabling, temperature and differences between test stations all add variation. Gage R&R puts a number on it.
Repeatability and reproducibility on an automated tester
| Component | Classic definition | What it means on ATE |
|---|---|---|
| Repeatability (equipment variation) | Variation when one operator measures the same part repeatedly with one gage | Variation when the same unit is re-tested on the same station: instrument noise, fixture contact, settling time |
| Reproducibility (appraiser variation) | Variation between operators | Variation between test stations, fixtures or nests, and between operators only where they load or connect the unit |
| Part variation | Real differences between the parts | Real differences between the units, which the test is meant to detect |
On a multi-station test line, reproducibility is usually the station-to-station difference. A product that passes on station 1 and fails on station 3 is a reproducibility problem, and it is one of the most common causes of retest loops.
How to run a gage R&R study on ATE
- Pick the measurements that matter. Focus on the tests with tight limits or a history of false failures, not every step in the sequence.
- Select about 10 units that span the real production range, including some near each limit. Ten identical, mid-range units make the tester look worse than it is.
- Choose the “appraisers”. Usually 2 or 3 test stations or fixtures. Include operators only if they influence the result by loading or connecting the unit.
- Test each unit 2 or 3 times on each station, in random order, re-seating the unit each time so fixture contact is part of the study.
- Record raw values, not just pass or fail. A study needs the measured numbers, so the test software should log them.
- Analyze with the ANOVA method in your statistics software and review %GRR, percent of tolerance and the number of distinct categories.
Ten units, three stations and three trials gives 90 measurements per test step, which an automated system can collect quickly.
How to read the results
| %GRR (of tolerance or study variation) | Common interpretation |
|---|---|
| Under 10% | Measurement system is generally acceptable |
| 10% to 30% | May be acceptable depending on the application, cost and risk; improvement is usually worthwhile |
| Over 30% | Not acceptable; the tester cannot reliably separate good from bad near the limits |
These are the widely used AIAG guidelines; your customer or industry may set different limits. Also check the number of distinct categories (ndc): a value of 5 or more means the system can resolve meaningful differences between parts.
Worked example
A tester measures an output voltage with limits of 4.90 V to 5.10 V, a tolerance of 0.20 V. The study finds a repeatability standard deviation of 0.0030 V and a station-to-station reproducibility of 0.00265 V.
- Combined gage R&R: √(0.0030² + 0.00265²) = √(0.00000900 + 0.00000702) = √0.00001602 ≈ 0.0040 V
- Six-sigma spread of the measurement system: 6 × 0.0040 = 0.024 V
- Percent of tolerance: 0.024 ÷ 0.20 × 100 = 12%
At 12%, the tester is usable but marginal, and nearly half the gage variation comes from differences between stations. Matching stations would be the first improvement to make.
How to improve a failing gage R&R
| If the main problem is… | Look at |
|---|---|
| Repeatability | Fixture contact and probe wear, instrument range and resolution, settling and averaging time, grounding and noise, temperature |
| Reproducibility between stations | Instrument calibration offsets, cable and fixture differences, station-specific software settings, compensation values |
| Both | Whether the limits are realistic for the measurement method, and whether the measurement belongs at this test step |
Many fixes are in the hardware: test fixture design, signal conditioning and shielding. Others are in the test software: longer settling, averaging, or per-station compensation that is itself verified.
Keeping a test system capable over its life
A gage R&R at acceptance is a snapshot. Fixtures wear, instruments drift and stations get repaired. To keep a tester trustworthy:
- Repeat the study after any major change: a new fixture, instrument replacement, software change to a measurement, or a move to a new site.
- Run a reference or “golden” unit at the start of each shift and chart the results to catch drift early.
- Track retest and false-failure rates by station; a rising rate on one station is often the first sign of a problem.
- Keep calibration records tied to each station and instrument.
- Include a gage R&R in the factory acceptance test of any new tester.
For quality and operations leaders
A capable tester protects two things at once: customers from escapes, and the plant from scrapping or retesting good product. If a line has a high retest rate, a gage R&R is often the fastest and cheapest way to find out whether the product or the tester is the problem.
Is your tester the problem?
Sciotex designs and builds custom ATE systems in the US since 1997, and can assess and improve existing testers on any hardware or language, including NI and LabVIEW systems.
Frequently asked questions
Gage R&R (gauge repeatability and reproducibility) is a study that measures how much of the variation in measured results comes from the measurement system itself, compared with the variation between the parts and the tolerance.
Under the widely used AIAG guidelines, below 10% is generally acceptable, 10% to 30% may be acceptable depending on the application, and above 30% is not acceptable.
Test about 10 units that span the production range, 2 or 3 times each on 2 or 3 test stations or fixtures, in random order, log the raw measured values, and analyze them with the ANOVA method.
On automated test equipment, reproducibility is mainly the difference between test stations or fixtures, plus operators only where they load or connect the unit.
Repeat it after any change that could affect the measurement, such as a new fixture, replaced instrument, measurement software change or relocation, and use daily reference-unit checks to catch drift in between.